EP0402249B1 - Wavelength stabilized semiconductor laser - Google Patents

Description

The subject of the present invention is a wavelength stabilized semiconductor laser.

It finds an application in optical telecommunications. It applies to all semiconductor lasers and in particular to lasers with single or double heterostructure in III-V alloy. It may, for example, but not exclusively, be lasers with GaAs, GaAlAs, InGaAs, GaAsP, InGaAsP, InP, etc.

In the field of optical telecommunications, there is a real need for laser sources offering high frequency stability. It is only on this condition that we will be able to implement radioelectricity techniques such as modulation, demodulation, etc.

Frequency stabilization of lasers in general and of semiconductor lasers in particular is a technique widely known today. Several solutions have been proposed: slaving the frequency to a resonance mode of a Fabry-Pérot standard or to a gas absorption line, synchronization on another laser, itself stabilized, etc.

To facilitate obtaining this stability, the laser to be stabilized is chosen or modified to already have a certain frequency stability. It is thus possible to operate by sorting double heterostructures in order to retain only the components whose alloy leads to the desired central frequency; it is also possible to reduce the natural emission width by lengthening the resonant cavity by means of an external auxiliary mirror (optical feedback); we can finally use a diffraction grating placed along the active area (laser called "DFB" for "Distributed Feed-Back" or "DBR" for "Distributed Bragg Reflection").

This state of the art is described for example in the article by Motoichi OHTSU entitled "Frequency Stabilization in Semi-Conductor Lasers" published in the journal "Optical and Quantum Electronics", vol. 20, (1988), pages 283-300.

However, these known techniques have drawbacks. The heterostructure sorting method leads to low manufacturing yields. The use of optical feedback by external auxiliary mirror poses difficult mechanical problems and leads to complex and cumbersome structures to implement. As for the lasers with networks distributed along the active area, they are obtained only at the cost of a complication of the manufacturing processes, due to epitaxial resumptions.

The object of the present invention is precisely to remedy all of these drawbacks. To this end, it offers a stabilization means which is particularly simple to implement and which avoids all the pitfalls of the prior art.

According to the invention, a crystal doped with rare earth ions is added to the laser cavity to stabilize. It is recalled in this connection that such ions have a ground state and a first excited electronic state. Due to the electric field created by the crystal lattice, these levels are broken down into sub-levels (Stark effect). The ions then have series of absorption lines corresponding to transitions from the ground state to the excited state and emission lines corresponding to reverse transitions. These two series of lines are slightly offset from each other, the emission lines being located at a slightly lower energy, therefore at a slightly longer wavelength.

We have already thought about using rare earths in semiconductor lasers. The article by W.T. TSANG et al. entitled "Observation of enhanced single longitudinal mode of operation in 1.5 µm GaInAsP erbium-doped semiconductor injection lasers" published in the journal "Applied Physics Letters", vol. 49, (25), December 22, 1986, pp. 1686-1688, describes a technique of introducing rare earth ions into the active area of the laser.

In reality, such a procedure leads to difficulties. First of all, it is difficult to introduce rare earths into the semiconductors (maximum concentration of the order of 10¹⁶-10¹⁸ cm⁻³) because they tend to precipitate there. We actually get a random multiple cavity, which gives a monofrequency operation but at an unpredictable frequency. Furthermore, the temperature instability observed by the authors W.T. TSANG et al. remains high: Δλ / ΔT = 1.1Å / ° C, (1Å = 10⁻¹⁰m.) which clearly shows that it is not the rare earth ions which stabilize the emission because the instability proper to these ions is much lower and around 10⁻²Å / ° C.

The inventor of the present invention has clearly shown (in an article published in the journal Journal of Applied Physics, vol. 66, 3952, 1989) that the rare earth ions enter only a low concentration in the semiconductor (case of Yb³⁺) or remaining outside of it (case of Er³⁺), the gains obtained at the frequency of the transition of the rare earth are negligible (10⁻¹ to 10⁻²cm⁻¹) compared to those of semiconductor (about 10²cm⁻¹). There is therefore hardly any modification of this gain by the addition of earth ions. rare.

An article by VAN DER ZIEL published in the journal Applied Physics Letters, vol. 50, 1313, 1987 confirmed that it was necessary to question the interpretations given previously in the article by W.T. TSANG.

The present invention overcomes all these drawbacks. It allows much higher concentrations of rare earth ions to be used than in the prior art (10²⁰ to 10²²cm⁻³), which makes it possible to considerably modify the gain of the semiconductor and to truly benefit from the great stability. in wavelength of the transition of the rare earth ion (instability of the order of 10⁻²Å / ° C).

This result is obtained thanks to the invention by using a crystal doped with rare earth ions placed outside the semiconductor. The rare earth ions are therefore no longer located in the semiconductor active medium.

According to the invention, the composition of the crystal and the nature of the ions are chosen so that one of the transitions of the rare earth ion (absorption and / or emission) falls into the spontaneous emission band of the semiconductor amplifying medium. . There is then a transfer of energy from the shortest wavelengths to the longest wavelengths, as will be better understood later. The transition of the rare earth ion happens to be privileged and it is on the wavelength of this transition that the laser will oscillate.

Specifically, the subject of the present invention is a semiconductor laser comprising an amplifying medium with a semiconductor junction and at least one mirror, the junction being joined to a current source and having an emission band. spontaneous, this laser being characterized by the fact that it further comprises, to stabilize it in wavelength, a crystal doped with rare earth ions having a transition falling in the spontaneous emission band of the amplifying medium, this crystal being placed between the mirror and the amplifying medium, the laser then emitting at a wavelength corresponding to the transition of the rare earth ion.

The value of the stabilized wavelength depends first of all on the rare earth ion used, and to a lesser extent on the crystal lattice, which imposes an electric field on the ions, thus displacing the levels by Stark effect. The intensity of the crystal field can thus vary by more than a factor of 3 between NdCl₃ and Y₃Al₅O₁₂: Nd for example.

• Sm²⁺ pour une longueur d'onde voisine de 0,65 µm à 0,7 µm,
• Nd³⁺, Yb³⁺ pour une longueur d'onde voisine de 0,81 µm à 1 µm,
• Tm²⁺ pour une longueur d'onde voisine de 1,2 µm,
• Er³⁺ pour une longueur d'onde voisine de 1,5 µm,
• Tm³⁺, Ho³⁺ pour une longueur d'onde voisine de 2 µm.

Thus, for the rare earth ion, we can use:

• Sm²⁺ for a wavelength close to 0.65 µm to 0.7 µm,
• Nd³⁺, Yb³⁺ for a wavelength close to 0.81 µm to 1 µm,
• Tm²⁺ for a wavelength close to 1.2 µm,
• Er³⁺ for a wavelength close to 1.5 µm,
• Tm³⁺, Ho³⁺ for a wavelength close to 2 µm.

As for crystal, a wide variety of materials is possible. We can cite: Y₃Al₅O₁₂, LaCl₃, LaF₃, LaP₅O₁₄, CaWO₄, LiYF₄, Na₅La (WO₄) ₄, LiLaP₄O₁₂, KLaP₄O₁₂, etc ...

The properties of crystals doped with rare earth ions are studied in an article by the inventor of the present invention, F. AUZEL, article entitled "A Scalar Crystal Field Parameter for Rare Earth Ions; Meaning and Application to Energy Transfer", published in the book "Energy Transfer Processes in Condensed Matter", edited by B. Di BARTOLO, (PLENNUM), 1984.

According to another advantageous arrangement of the invention, the crystal is given a shape such that it produces a mode selection. By giving the stabilizing crystal the shape of a thin blade capable of playing the role of Fabry-Pérot standard, this standard may have only one longitudinal mode in the atomic line. In other words, the free spectral interval, conventionally given by the formula c / 2nL where c is the speed of light, n the index of the crystal and L the thickness of the plate, must be close to or greater at the width ΔF of the emission line of the rare earth ion.

We will therefore endeavor to take a thickness L greater than or equal to c / 2n ΔF.

For n = 1.5, ΔF = 1cm⁻¹, we obtain a thickness L of 1 mm.

According to another advantageous arrangement of the invention, several crystals of different composition doped with the same rare earth are used. For the reasons mentioned above, several preferred wavelengths will be obtained. The laser will then be multi-wavelength and find an application in optical multiplexing.

• la figure 1 montre schématiquement un spectre d'absorption-émission d'un ion terre rare dans un cristal,
• la figure 2 montre la courbe de gain d'un milieu aplificateur semi-conducteur et celle d'un ensemble composite comprenant un amplificateur semi-conducteur et un cristal dopé stabilisateur,
• la figure 3 illustre un premier mode de réalisation d'un laser conforme à l'invention,
• la figure 4 illustre un deuxième mode de réalisation d'un laser conforme à l'invention,
• la figure 5 illustre un troisième mode de réalisation d'un laser conforme à l'invention.

In any case, the characteristics of the invention will appear better in the light of the description which follows. This description relates to nonlimiting examples and refers to the attached drawings in which:

• FIG. 1 schematically shows an absorption-emission spectrum of a rare earth ion in a crystal,
• FIG. 2 shows the gain curve of a semiconductor amplifier medium and that of a composite assembly comprising a semiconductor amplifier and a stabilized doped crystal,
• FIG. 3 illustrates a first embodiment of a laser according to the invention,
• FIG. 4 illustrates a second embodiment of a laser according to the invention,
• FIG. 5 illustrates a third embodiment of a laser according to the invention.

FIG. 1 shows schematically the absorption spectrum (A) of a crystal doped with rare earth ions and the emission spectrum (E). The abscissa axis corresponds to the wavelength. The absorption peaks correspond to the different atomic transitions from the ground state (separated by the Stark effect) to the first excited electronic state (also separated by the Stark effect) and the emission peaks to the transitions from the excited state to the ground state. The shift towards the long wavelengths of the emission compared to the absorption is traditional (Stokes effect).

A crystal thus doped, placed in the cavity of a laser, will modify the gain of the amplifying medium according to the diagram in FIG. 2. The gain curve G represents the variations in gain as a function of the wavelength for a medium classic semiconductor amplifier. The presence of the doped crystal decreases the gain in the absorption zone and increases it in the emission zone to give a curve G ′ which is, roughly speaking, below the curve G for low wavelengths and above G for the longer wavelengths.

If the resonator has losses P, oscillation will only be possible for the wavelength where the gain will outweigh these losses. The oscillation will thus occur at the wavelength λ o, which is no longer defined by the top of the curve. gain of the semiconductor amplifying medium, with all the fluctuations which may result therefrom, but by the emission line specific to the rare earth ion, with the stability that is known to atomic lines.

The invention is not limited to operation on an emission line such as the line with λo shown, but extends to the case of any transition of the rare earth ion. At low excitation levels, even the "negative" part of the gain curve can be used to stabilize the laser.

Figures 3 to 5 show some embodiments of lasers according to the invention.

The laser illustrated in FIG. 3 comprises an amplifying medium 10 in the form of a double heterostructure with PN junction referenced 12. The semiconductor structure is supplied with current by a source 14. The front face 16 of the heterostructure is cleaved. The rear face is covered with an anti-reflection layer 18. A thin plate 20 of crystal doped with rare earth ions is placed between the anti-reflection layer 18 and a spherical mirror 22. The coherent emission stabilized in length wave is carried out from the front (reference 23).

In a more compact variant illustrated in FIG. 4, the doped crystalline thin plate 20 is pressed against the anti-reflection layer 18 and a reflective layer 24 is deposited on the rear face of the blade.

The variant illustrated in FIG. 5 differs from the previous ones by the use of an optical fiber 26 disposed between the crystal plate 20 and the reflective layer 24. This fiber is advantageously self-focusing. In this figure, there is also shown a second blade 20 ′ made of a crystal different from that of the blade 20 but doped with the same ions.

By way of nonlimiting example, a laser according to the invention will be described, emitting in a single longitudinal mode ("single mode" or "single frequency") stabilized at 1.5335 μm.

The amplifying medium consists of a double heterostructure in quaternary alloy whose active zone is in InGaAsP. The natural emission of such a structure is between 1.480 and 1.525 µm depending on the composition. The structure is stabilized by a crystal of LiYF₄ doped with erbium, of formula LiY _1-x Er _X F₄ where x is between 10⁻⁴ and 1. The value of x depends both on the power available in the cavity , the thickness of the crystal plate and the concentration of Er³⁺ ions. The higher the power, the more x can deviate from the unit and the smaller the thickness of the crystal can be chosen.

où
   T désigne la durée de vie du niveau excité de la terre rare et est pris égal à 10 ms,
   F est la fréquence, soit 2x10¹⁴s⁻¹ pour une longueur d'onde de 1,54 µm,
   s est la section efficace d'absorption et est prise égale à 10⁻²⁰cm⁻²,
   φ est le flux optique incident sur le cristal en Wcm⁻²,
   N est la densité d'ions,
   L est l'épaisseur du cristal stabilisateur,
   h est la constante de Planck,
   r est le rendement quantique qui peut être pris égal à l'unité.
Le second membre de l'inégalité est donc sensiblement égal à 6,6x10²² (Wcm⁻⁴).

• a) En prenant x=1, on a N=1,4x10²²cm⁻³ ; en conséquence, L doit être supérieur à 5Wcm⁻¹.
  Pour un laser ayant une puissance de 5mW dans la cavité, avec une zone active de 50 µmx0,1 µm, soit 5.10⁻⁸( µm)², le flux sera de 10⁵W/cm². L'épaisseur L du cristal devra donc être supérieure à 5.10⁻⁵cm, soit supérieure à un demi micromètre.
  Cet exemple montre que si le cristal stabilisateur a une grande concentration en ions de terre rare, il peut être très mince (0,5 µm) et constitué par exemple d'une couche disposée sur la face anti-reflet de l'hétérostructure. Mais la qualité du matériau ainsi déposé peut laisser à désirer et le rendement quantique diminuer.
  On peut donc préférer choisir une concentration en ions plus faible (x inférieur à 1).
• b) A l'autre extrémité de la plage offerte à x on peut choisir x=0,001, ce qui correspond à N=1,4x10¹⁹cm⁻³ et, avec les mêmes hypothèses sur la puissance, l'épaisseur L doit alors être supérieure à 100 µm.
  Si le flux lumineux correspond à une puissance 10 fois plus faible, l'épaisseur L devient au minimum égale à 1000 µm.
  Si l'on veut simultanément que la lame cristalline agisse comme sélecteur de mode longitudinal, il conviendra de ne pas lui donner une épaisseur supérieure à la valeur qui donne à l'intervalle spectral libre la valeur de la largeur de raie d'émission, à savoir sensiblement 3,3 cm⁻¹. On a vu que l'épaisseur limite est de l'ordre du millimètre.
• c) Entre ces deux extrêmes (x=1 et x=0,001), on pourra prendre par exemple x=0,1, soit N=1,4x10²¹cm⁻³, ce qui donne une épaisseur de L=0,3 mm.

More precisely, in the case of a uniform absorption in the stabilizing crystal, we have the approximate relation: $$\text{φLN>}\frac{\text{1}}{\text{2T}}\frac{\text{hF}}{\text{sr}}$$

or
T designates the lifetime of the excited level of the rare earth and is taken equal to 10 ms,
F is the frequency, i.e. 2x10¹⁴s⁻¹ for a wavelength of 1.54 µm,
s is the effective absorption section and is taken equal to 10⁻²⁰cm⁻²,
φ is the incident optical flux on the crystal in Wcm⁻²,
N is the density of ions,
L is the thickness of the stabilizing crystal,
h is Planck's constant,
r is the quantum yield which can be taken equal to unity.
The second member of the inequality is therefore substantially equal to 6.6x10²² (Wcm⁻⁴).

• a) Taking x = 1, we have N = 1.4x10²²cm⁻³; consequently, L must be greater than 5Wcm⁻¹.
  For a laser with a power of 5mW in the cavity, with an active area of 50 µmx0.1 µm, or 5.10⁻⁸ (µm) ², the flux will be 10⁵W / cm². The thickness L of the crystal must therefore be greater than 5.10⁻⁵cm, ie greater than half a micrometer.
  This example shows that if the stabilizing crystal has a large concentration of rare earth ions, it can be very thin (0.5 μm) and consists for example of a layer placed on the anti-reflection face of the heterostructure. However, the quality of the material thus deposited may leave something to be desired and the quantum yield may decrease.
  We may therefore prefer to choose a lower ion concentration (x less than 1).
• b) At the other end of the range offered to x we can choose x = 0.001, which corresponds to N = 1.4x10¹⁹cm⁻³ and, with the same assumptions on the power, the thickness L must then be greater than 100 µm.
  If the luminous flux corresponds to a power 10 times lower, the thickness L becomes at least equal to 1000 μm.
  If we simultaneously want the crystal plate to act as a longitudinal mode selector, it should not be given a thickness greater than the value which gives the free spectral interval the value of the emission line width, at know substantially 3.3 cm⁻¹. We have seen that the limit thickness is of the order of a millimeter.
• c) Between these two extremes (x = 1 and x = 0.001), we can take for example x = 0.1, or N = 1.4x10²¹cm⁻³, which gives a thickness of L = 0.3 mm.

Such a thin blade can be obtained by polishing. One reflective gold layer is deposited on one of its faces intended to play the role of mirror with high reflectivity at the wavelength of 1.5335 μm. The rear face of the double heterostructure in InGaAsP is treated with anti-reflection by a layer of SiO producing a reflection of less than 10⁻². The assembly according to FIG. 4 is carried out using thin adhesives so as not to destroy the parallelism of the surfaces. Adhesives of the cyanoacrylate or epoxy type called "for IR optics" are very suitable.

Claims (11)

1. Semiconductor laser including an amplifier medium (10) with a semiconductive junction (12) and at least one mirror (22), the junction (12) being joined to a current source (14) and having one spontaneous emission band, characterized in that to wavelength-stabilize the same, it further includes a crystal doped with rare earth ions having a transition within the spontaneous emission band of the amplifier medium, said crystal (20) being disposed between the mirror (22) and the amplifier medium (10), the laser then transmitting on a wavelength corresponding to the transition of the rare earth ion.
2. Semiconductor laser according to claim 1, characterized in that the rare earth ions are selected from the group constituted by:

   - Sm²⁺ for a wavelength of between 0.65 µm and 0.7 µm,

   - Nd³⁺, Yb³⁺ for a wavelength of between 0.81 µm and 1 µm,

   - Tm²⁺ for a wavelength of about 1.2 µm,

   - Er³⁺ for a wavelength of about 1.5 µm,

   - Tm³⁺, Ho³⁺ for a wavelength of about 2 µm.
3. Semiconductor laser according to claim 1, characterized in that the crystal is taken from the group constituted by LiYF₃, Y₃ Al₅ O₁₂, LaCl₃, LaF₃, LaP₅ O₁₄, CaWO₄, LiYF₄, Na₅La(WO₄)₄. LiLaP₄ O₁₂, KLaP₄ O₁₂.
4. Semiconductor laser according to claims 2 and 3, characterized in that the formula of the doped crystal is LiY_1-xEr_x F₄ where x is between 0 and 1 (lower limit excluded).
5. Semiconductor laser according to claim 1, characterized in that the doped crystal has the shape of a thin plate (20) cut from a monocrystal, this thin plate constituting a Fabry-Pérot standard having a thickness less than the value giving this standard a free spectral interval of about the width of the emission line of the rare earth ion.
6. Semiconductor laser according to claim 1, characterized in that the doped crystal has the shape of a thin plate.
7. Semiconductor laser according to claim 1, characterized in that the amplifier medium (10) includes a heterostructure having one split face (16) and one face covered with an anti-reflection coating (18), the crystal (20) being disposed between the coating (18) and the mirror (22).
8. Semiconductor laser according to claim 7, characterized in that the crystal (20) is in contact with the anti-reflection coating (18).
9. Semiconductor laser according to claim 7, characterized in that the doped crystal (20) is in contact with the mirror (24).
10. Semiconductor laser according to claim 8, characterized in that it further includes an optical fibre (26) disposed between the doped crystal (20) and the mirror (24).
11. Semiconductor laser according to claim 1, characterized in that it includes several crystals (20,20') having a different chemical nature, but containing the same rare earth ions, the laser then being stabilized on several wavelengths.

Images